November 20, 2026

AI Course Creation Tools in 2026: What Changed for UK L&D Teams

A close up of a computer circuit board

UK businesses adopted AI faster than most forecasts had allowed for. The Office for National Statistics puts adoption among firms with 10 or more employees at 35% in June 2026, up from roughly 12% in late 2023. Large language models lead every other category at 18%.

Depth is a different story. The average adopting business runs 1.6 AI technologies, up from 1.4 three years ago, and only one adopter in ten describes its use as extensive. Wide, but thin.

Training content sits right in that gap. Plenty of L&D teams have a ChatGPT tab open. Far fewer have changed how a course gets built.

“AI-powered” has stopped being a useful filter

Almost every authoring tool now advertises AI somewhere above the fold, which makes the label difficult to compare on. A narrower question works better: where in the production cycle does the AI remove work?

Course production stalls in fairly predictable places. Somebody has to read the source material and decide what matters. Then, somebody writes the first draft. Then assessment questions. Then narration. Then a version for the teams who don’t work in English. A tool whose AI capabilities don’t reach at least one of those stages is unlikely to change how quickly training is delivered.

The slide course was the last format to receive it

iSpring is worth looking at here, partly because its 2026 releases arrived in sequence rather than as one bundle, and each targeted a different stage of production.

In May, an AI course generator was added to iSpring LMS and iSpring Cloud AI. An author provides a brief description and a few source files; the tool returns an outline, draft text, and assessment questions, edited in the browser as a scrollable course.

In June, ElevenLabs voices were added to iSpring’s narration, enabling voiceovers to be generated straight from the script. A course that would otherwise have waited for a recording session or a voice artist can now be narrated the same afternoon.

The format still outside that scope was the one most corporate training runs on: the narrated slide deck, published as a SCORM package. That changed in August, when iSpring extended AI generation to slide courses in iSpring Suite AI.

The sequence matters more than the generation itself. An author describes the course in plain language, then uploads up to five documents for the AI to work from. Specifications. Manuals. Process notes nobody has opened since the last audit.

Nothing gets built yet. First, the AI proposes the course parameters: topic, course goal, target audience, learning objectives, and writing style. An outline follows, carrying a course title, description, and module structure, where modules can be rewritten, reordered, duplicated, or cut.

That order is a design decision worth noticing. The difficult step in course creation has rarely been opening the editor. It is working out what the course is for, who is sitting in front of it, and what someone should be able to do afterwards. A generator that goes straight to writing slides leaves all of that to the author. Brief first, structure second, slides third is the order an instructional designer would work in anyway, and building it into the tool means a subject matter expert reaches a defensible course without having been taught the method.

The language is set at the outline stage too, which means a course for a Dutch or Spanish team can be generated in that language rather than written in English and translated afterwards.

Then the slides. The draft opens as a presentation: title slide, content slides, section dividers, and closing slide, with speaker notes throughout. Authors decide at the settings stage whether AI writes the voiceover script. Ask for it, and the narration goes into the notes, leaving less text on the slide itself, which suits a course that’s meant to be watched rather than read.

The outline is saved inside the presentation file, so an author can reopen it from the iSpring Suite AI tab and generate it again. A new draft either replaces the existing slides or opens as a separate presentation, which lets two treatments of the same structure sit side by side.

Interactions, quizzes, and visual design remain the author’s work. iSpring says so plainly rather than overselling the output, which is a reasonable position to take. The result is a first draft in PowerPoint with a defensible structure behind it.

Who ends up building the course

The consequential shift sits outside the feature list. It concerns who can now start building courses.

Instructional designers have been the bottleneck for a defensible reason: building a course from nothing is a real skill. Change the starting point to a generated structure, and the job changes shape. A subject matter expert who writes training twice a year now has something to work from. Design time moves toward refinement.

For small teams, that is the entire value. One or two people carrying training alongside three other jobs do not need a better blank slide. They need a first version to react to.

Where a pilot should focus

Teams running a trial tend to learn the most from three things.

Start with the material you already have. Use manuals and process documents as input rather than a tidy brief written for the occasion, because real projects rarely start that way.

Then push the draft around. Reorder modules, rewrite a section, and regenerate against the same outline. The value shows up in how far the structure bends, not in how the first version reads.

Finally, publish something. Send a SCORM or xAPI package to your LMS and watch how it behaves there. If you train across markets, run a course through translation and have a native speaker read it, less because the output will be wrong and more because terminology is worth a local eye.

Authoring software has spent this year sorting out which parts of course production AI can absorb. On the evidence of the last three releases, the answer is the mechanical stretch: reading the source material, proposing a structure, filling in a first version of the text. That is a real amount of work to hand over, and it is the part that most often stops a course from getting started at all.

What has not moved is the judgement. Deciding what belongs in the course, whether the examples ring true for the people taking it, and whether any of it changes how they work afterwards is still the job. The difference is that it is now the entire job, rather than the thing you get to after three days of building slides.

Teams that want to test the generator with their own source material can try it with a 14-day trial of iSpring Suite AI.